Negotiating a Microsoft Power Platform deal across six meters

The Power Platform bills across six independent meters, and the Power Apps per app plan reached end of sale on January 2, 2026. On a representative 7,500 seat estate, a blanket renewal runs about $0.56 million a year above a targeted deal, roughly 36 percent.

Prepared by Redress Compliance · June 2026 · Representative Power Platform estate scenario (benchmark scenario, not a quote).

1. How is a Power Platform deal actually priced?

A Power Platform deal is priced on six meters that share nothing but a logo. You negotiate each one separately, because Microsoft prices each one separately and the account team will bundle them only when bundling favors Microsoft.

The first move is to read the public price book and write it down. Every benchmark you set later anchors on these numbers, sourced from the Microsoft Power Platform pricing page.

Three of these meters did not exist in their current form two years ago. Treat any quote built on last year's SKUs as a draft, not an offer.

2. Power BI: where does the money leak across Pro, PPU, and Fabric?

Power BI leaks money when viewers are licensed as authors. Most of a Power BI population reads dashboards. A smaller group builds them. Licensing both groups the same way is the most common Power BI overspend we see.

• Power BI Pro at $14 per user per month suits authors and collaborators who share content.
• Premium Per User at $24 per user per month adds paginated reports, larger models, and higher refresh rates for power authors.
• Fabric capacity (an F SKU) lets unlimited viewers consume content with no per user license once an F64 or larger capacity is in place.

The non obvious mechanic: the read only viewer break point. Above roughly 250 to 300 viewers, a Fabric F64 capacity is usually cheaper than buying every viewer a Pro seat. Below that, per user wins. Model both before you commit.

3. Power Apps: per user or pay as you go after the per app retirement?

Power Apps is where the per user versus per app trap lives, and it changed in January 2026. Microsoft moved the per app plan to end of sale for new buyers on January 2, 2026. Existing Enterprise Agreement and CSP customers can still renew and add per app at true up, which is itself a lever.

The replacement for new buyers is Power Apps Premium at $20 per user per month, or the pay as you go meter at $10 per active user per app per month. The trap is licensing every app user at the Premium rate when most of them touch one app.

On the representative estate, 4,200 app users were quoted on Premium. Only 900 are multi app makers. The other 3,300 run a single app and bill far less on the pay as you go meter.

The buyer side move: count how many apps each user actually runs, map most users to pay as you go, and reserve Premium for heavy makers only. Then hold the per app renewal right if you are an existing EA customer, because it is a price you can keep that new buyers cannot.

4. Power Automate: do you license the users or the processes?

Power Automate is priced two ways, and choosing wrong doubles the bill. License the users when humans trigger flows. License the process when a flow runs unattended on a schedule for the whole business.

• Power Automate Premium at $15 per user per month covers users running premium connectors and desktop flows under their own identity.
• Hosted process at $215 per bot per month covers unattended automation that serves many users from one Microsoft hosted bot.

The non obvious mechanic: a single shared automation that serves 200 users needs one process license, not 200 user licenses. Inventory your flows by trigger type before you accept a per user quote for automations no human launches.

5. Power Pages: how do authenticated and anonymous users bill?

Power Pages bills external users, not your staff, and the two user types bill differently. Authenticated users come in per 100 user packs at $200 per month, tiered down as volume rises. Anonymous users come in separate capacity packs.

The trap is buying authenticated capacity for traffic that never signs in. Audit your portal analytics first. Most public facing pages are dominated by anonymous reads, which carry a lower meter than the authenticated packs the account team will quote by default.

• Authenticated: size to peak concurrent signed in users, not registered accounts.
• Anonymous: size to monthly page views against the capacity pack allowance.
• Tiering: the per user rate falls as packs stack, so consolidate sites onto shared capacity where the data model allows.

6. Copilot Studio: what does credit based pricing really cost?

Copilot Studio bills on credits, and credits are the easiest meter to over commit. A pack is $200 per month for 25,000 credits. Agent messages, generative answers, and AI Builder actions all draw from the same credit pool.

The trap, and it repeats, is committing credit packs before any measured consumption exists. AI Builder credit commitments used to be a separate line. They now flow through Copilot Studio credits, so a team that pre buys for a pilot that never scales pays for empty capacity all term.

The buyer side move: run agents on pay as you go credit metering through the pilot, measure the real monthly draw, then commit packs only to the consumption you have actually observed. Commitment buys a small unit discount, never enough to justify paying for credits you do not burn.

7. Dataverse: how do you cap capacity before it bills you?

Dataverse is the silent overage. Database capacity beyond the allowance included with your seats bills at $40 per gigabyte per month, and it accrues quietly as solutions grow. Two in three estates we benchmark carry unmanaged Dataverse overage at renewal.

The non obvious mechanic: file and log storage are cheaper meters than database storage. Moving attachments to file capacity and audit data to log capacity, then archiving stale tables, can cut a database overage line by half before you negotiate the rate.

• Audit capacity in the admin center before the renewal, not after the true up.
• Reclassify storage so attachments and logs leave the expensive database meter.
• Forecast capacity from real solution designs, not from a vendor rule of thumb.

8. What does the whole estate look like, blanket versus targeted?

Put the six meters together and the pattern is consistent. The blanket renewal the account team quotes carries roughly a third more than a targeted deal that maps each population to its lowest sufficient meter.

On the representative 7,500 seat estate, the difference is $559,800 a year. The summary below ties every meter back to the sections above.

Benchmark ranges: Redress Compliance advisory engagement file, 2024 to 2025.

9. How do you build an entitlement baseline that survives Microsoft scrutiny?

You cannot negotiate a count you cannot defend. The entitlement baseline is the single artifact that decides whether the renewal opens on your numbers or Microsoft's. Build it before you ask for a quote.

A baseline that survives scrutiny reconciles five sources and ties every license to a named consumption signal. When the account team challenges a number, you answer with telemetry, not opinion.

• Admin center inventory: environments, apps, flows, and Dataverse capacity, exported with dates.
• Per app usage telemetry: active users per app over a 90 day window, the evidence for per app mapping.
• Premium connector and desktop flow logs: who actually triggers premium automation.
• Copilot Studio credit draw: measured monthly consumption, not the pilot's pre buy.
• Contract entitlements: what your current EA, CSP, or MCA already grants, including any held per app renewal right.

The non obvious mechanic: Microsoft prices from your future state, you price from your measured state. The baseline forces the conversation onto observed usage, where your numbers are defensible and theirs are projections.

10. Which five contract clauses decide whether the commitment protects the budget?

A discount is a number on page one. The clauses on page twelve decide whether that number holds when your estate changes. Five clauses do most of the work on a Power Platform commitment.

The clause buyers forget is meter substitution. A three year lock to Premium when your population is drifting to single app use means you pay the structural premium for the whole term, whatever the discount.

11. What discount benchmarks hold across renewal and exit?

Discount benchmarks depend entirely on your alternative. A flat renewal with no credible move earns a flat renewal discount. A renewal with a real exit on the table earns materially more. The ranges below come from our engagement file.

The lesson holds across every meter. The discount is a function of your willingness to walk on at least part of the estate. Power Apps and Power Automate have real alternatives. Use them to set the band.

12. How do you build a BATNA and a side letter against Microsoft tactics?

Your best alternative is the only thing that moves a Microsoft discount. A BATNA on Power Platform is rarely all or nothing. It is credible substitution on the meters where alternatives exist, priced and ready to name.

• Power BI: a competing analytics platform for the viewer population caps the Fabric capacity ask.
• Power Automate: a standalone automation tool for shared processes caps the hosted process count.
• Power Apps: a competing low code platform for net new builds, even partial, sets a ceiling on the Premium expansion.

The tactics to expect are standard. The account team will bundle Copilot and Dataverse into the apps deal, quote blanket Premium as the simple option, and time the offer to your urgency rather than theirs. Counter each one in writing.

The side letter language we use protects the structure: a documented right to hold the per app renewal rate, to substitute meters at anniversary, and to true down capacity. Get it in a signed side letter, not a verbal assurance from the account team, because account teams rotate and verbal terms do not survive the rotation.

13. The eleven move buyer side framework

Pulled together, the negotiation runs on eleven moves. Run them in order and the discount conversation becomes the last step, not the first.

1. Build the entitlement baseline from five reconciled sources before you ask for a quote.
2. Count apps per user and map most users to pay as you go.
3. Reserve Premium for heavy multi app makers only.
4. Hold the per app renewal right if you are an existing EA or CSP customer.
5. Set the Power BI viewer break point and choose Fabric capacity above it.
6. Inventory flows by trigger and license shared automations per process.
7. Audit Dataverse capacity and reclassify storage off the database meter.
8. Meter Copilot Studio on consumption through the pilot before committing packs.
9. Secure the five clauses: price protection, quantity reduction, meter substitution, capacity true down, discount on adds.
10. Name a credible BATNA on at least two meters and price it.
11. Time the close to Microsoft fiscal year end, June 30, and put the structure in a signed side letter.

Where is the common advice on Power Platform negotiation wrong?

The standard reseller pitch is to standardize on Power Apps Premium for simplicity and chase the deepest blanket discount. We disagree.

In the Power Platform estates Redress Compliance benchmarked in 2024 to 2025, the blanket Premium quote ran roughly 30 to 50 percent above a targeted plan on the app meter alone. No realistic discount closes that gap. Standardization is sold as administrative ease, but it is a structural premium you pay every month of the term.

The buyer side move is to license the population, not the platform. Map each user to the lowest sufficient meter, govern Dataverse and credits, and let structure carry the saving, not the discount line.